290 CORALS AND CORAL ISLANDS. 
chain of atolls is continued northward in the Ratack and 
Ralick Groups (or the Marshall Islands), making in all a 
chain over 1,200 miles long; and, adding the concordant 
Kllice Islands on the south, and extending the Ratack line to 
Gaspar Rico, its northern outlier, the chain is nearly 2,000 
miles long. Nothing in the direction of the long range, ex- 
cepting local shapings of some of the points about the atolls, 
‘can be attributed to the Pacific currents. Moreover, none of 
, the diversified forms of atolls have any sufficient explanation 
~ in the drift process. 
Dr. Guppy urges the idea that banks of reef after reach- 
ing the surface are converted into atolls by means of the 
marine currents.’ In his last paper he speaks of the conver- 
sion after the island has been thrown up by the waves; in the 
earlier he appears to speak of elevation of some kind as 
necessary ; but the currents would act the same, whatever 
the means of reaching the surface. In the groups of atolls 
above referred to, neither the forms of the atolls nor the 
currents favor such an hypothesis. In the Paumotus, there 
are no currents of strength enough for work of this kind, 
as shown on page 296. He speaks of Keeling Atoll as a 
horse-shoe or crescentic island, and states that such forms are 
common among atolls. But the Keeling Atoll, according to 
the best maps, is an ordinary atoll, the lagoon nearly encircled 
with reef, and the application of such terms to it 1s wholly 
misleading. The shaping of a reef by the horse-shoe method 
of drifting and making an eddy to leeward which he describes, 
is wholly inapplicable to the ordinary atolls of the ocean; for 
there are no means for producing the result. 
d. Further, drifting by currents may make beaches and 
1 On the Solomon Islands, Proceedings of the Royal Society of Edinburgh, 
1885-1886, xiii. 857; On the Keeling Atoll, Nature, January, 1889, xxxix. 236. 

